/[webpac]/trunk/all2xml.pl
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Contents of /trunk/all2xml.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 74 - (show annotations)
Sat Jul 5 22:37:30 2003 UTC (16 years, 4 months ago) by dpavlin
File MIME type: text/plain
File size: 12774 byte(s)
support for new feed format which have decimal number of field, semicolumn
and space at beginning of each line (like: 0: data)

1 #!/usr/bin/perl -w
2
3 use strict;
4 use OpenIsis;
5 use Getopt::Std;
6 use Data::Dumper;
7 use XML::Simple;
8 use Text::Unaccent 1.02; # 1.01 won't compile on my platform,
9 use Text::Iconv;
10 use Config::IniFiles;
11 use Encode;
12
13 $|=1;
14
15 my $config_file = $0;
16 $config_file =~ s/\.pl$/.conf/;
17 die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file);
18
19 my $config;
20
21 #use index_DBI; # default DBI module for index
22 use index_DBI_cache; # faster DBI module using memory cache
23 my $index;
24
25 my %opts;
26
27 # usage:
28 # -d directory name
29 # -m multiple directories
30 # -q quiet
31 # -s run swish
32
33 getopts('d:m:qs', \%opts);
34
35 my $path; # this is name of database
36
37 Text::Iconv->raise_error(0); # Conversion errors don't raise exceptions
38
39 # this is encoding of all files on disk, including import_xml/*.xml file and
40 # filter/*.pm files! It will be used to store strings in perl internally!
41 my $codepage = 'ISO-8859-2';
42
43 my $utf2cp = Text::Iconv->new('UTF-8',$codepage);
44 # this function will convert data from XML files to local encoding
45 sub x {
46 return $utf2cp->convert($_[0]);
47 }
48
49 # decode isis/excel or other import codepage
50 my $import2cp;
51
52 # outgoing xml must be in UTF-8
53 my $cp2utf = Text::Iconv->new($codepage,'UTF-8');
54
55 # mapping between data type and tag which specify
56 # format in XML file
57 my %type2tag = (
58 'isis' => 'isis',
59 'excel' => 'column',
60 'marc' => 'marc',
61 'feed' => 'feed'
62 );
63
64 sub data2xml {
65
66 use xmlify;
67
68 my $type = shift @_;
69 my $row = shift @_;
70 my $add_xml = shift @_;
71 # needed to read values from configuration file
72 my $cfg = shift @_;
73 my $database = shift @_;
74
75 my $xml;
76
77 use parse_format;
78
79 my $html = ""; # html formatted display output
80
81 my %field_usage; # counter for usage of each field
82
83 # sort subrouting using order="" attribute
84 sub by_order {
85 return 0 if (! $config->{indexer}->{$a}->{order});
86 return 0 if (! $config->{indexer}->{$b}->{order});
87
88 return $config->{indexer}->{$a}->{order} <=>
89 $config->{indexer}->{$b}->{order} ;
90 }
91
92 foreach my $field (sort by_order keys %{$config->{indexer}}) {
93
94 $field=x($field);
95
96 $field_usage{$field}++;
97
98 my $swish_data = "";
99 my $display_data = "";
100 my $line_delimiter;
101
102 my ($swish,$display);
103
104 my $tag = $type2tag{$type} || die "can't find which tag to use for type $type";
105 foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) {
106
107 my $format = x($x->{content});
108 my $delimiter = x($x->{delimiter}) || ' ';
109
110 my $repeat_off = 0; # repeatable offset
111
112 my ($s,$d,$i) = (1,1,0); # swish, display default
113 $s = 0 if (lc($x->{type}) eq "display");
114 $d = 0 if (lc($x->{type}) eq "swish");
115 ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
116
117 # what will separate last line from this one?
118 if ($display_data && $x->{append} && $x->{append} eq "1") {
119 $line_delimiter = ' ';
120 } elsif ($display_data) {
121 $line_delimiter = '<br/>';
122 }
123
124 # init vars so that we go into while...
125 ($swish,$display) = (1,1);
126
127 if ($swish || $display) {
128 ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
129 # filter="name" ; filter this field through
130 # filter/[name].pm
131 my $filter = $x->{filter};
132 if ($filter) {
133 require "filter/".$filter.".pm";
134 }
135 # type="swish" ; field for swish
136 if ($s && $swish) {
137 if ($filter) {
138 no strict 'refs';
139 $swish_data .= join(" ",&$filter($swish));
140 } else {
141 $swish_data .= $swish;
142 }
143 }
144
145 # type="display" ; field for display
146 if ($d && $display) {
147 if ($line_delimiter && $display_data) {
148 $display_data .= $line_delimiter;
149 undef $line_delimiter;
150 }
151 if ($filter) {
152 no strict 'refs';
153 $display_data .= join($delimiter,&$filter($display));
154 } else {
155 if ($display_data) {
156 $display_data .= $delimiter.$display;
157 } else {
158 $display_data .= $display;
159 }
160 }
161 }
162
163 # type="index" ; insert into index
164 if ($i && $display) {
165 my $index_data = $display;
166 if ($filter) {
167 no strict 'refs';
168 foreach my $d (&$filter($index_data)) {
169 $index->insert($field, $d, $path);
170 }
171 } else {
172 $index->insert($field, $index_data, $path);
173 }
174 }
175 }
176 }
177
178 # now try to parse variables from configuration file
179 foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) {
180
181 my $delimiter = x($x->{delimiter}) || ' ';
182 my $val = $cfg->val($database, x($x->{content}));
183
184 my ($s,$d,$i) = (1,1,0); # swish, display default
185 $s = 0 if (lc($x->{type}) eq "display");
186 $d = 0 if (lc($x->{type}) eq "swish");
187 ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
188
189 if ($val) {
190 $display_data .= $delimiter.$val if ($d);
191 $swish_data .= $val if ($s);
192 $index->insert($field, $val, $path) if ($i);
193 }
194
195 }
196
197
198 if ($display_data) {
199
200 if ($field eq "headline") {
201 $xml .= xmlify("headline", $display_data);
202 } else {
203
204 # find field name (signular, plural)
205 my $field_name = "";
206 if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) {
207 $field_name = $config->{indexer}->{$field}->{name_singular}."#-#";
208 } elsif ($config->{indexer}->{$field}->{name_plural}) {
209 $field_name = $config->{indexer}->{$field}->{name_plural}."#-#";
210 } elsif ($config->{indexer}->{$field}->{name}) {
211 $field_name = $config->{indexer}->{$field}->{name}."#-#";
212 } else {
213 print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
214 }
215 if ($field_name) {
216 $html .= x($field_name);
217 }
218 $html .= $display_data."###\n";
219 }
220 }
221 if ($swish_data) {
222 # remove extra spaces
223 $swish_data =~ s/ +/ /g;
224 $swish_data =~ s/ +$//g;
225
226 $xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
227 }
228
229
230 }
231
232 # dump formatted output in <html>
233 if ($html) {
234 $xml .= xmlify("html",$html);
235 }
236
237 if ($xml) {
238 $xml .= $add_xml if ($add_xml);
239 return "<xml>\n$xml</xml>\n";
240 } else {
241 return;
242 }
243 }
244
245 ##########################################################################
246
247 # read configuration for this script
248 my $cfg = new Config::IniFiles( -file => $config_file );
249
250 # read global.conf configuration
251 my $cfg_global = new Config::IniFiles( -file => 'global.conf' );
252
253 # open index
254 $index = new index_DBI(
255 $cfg_global->val('global', 'dbi_dbd'),
256 $cfg_global->val('global', 'dbi_dsn'),
257 $cfg_global->val('global', 'dbi_user'),
258 $cfg_global->val('global', 'dbi_passwd') || '',
259 );
260
261 foreach my $database ($cfg->Sections) {
262
263 my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
264 my $add_xml = $cfg -> val($database, 'xml'); # optional
265
266 print STDERR "reading ./import_xml/$type.xml\n";
267
268 # extract just type basic
269 my $type_base = $type;
270 $type_base =~ s/_.+$//g;
271
272 $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1);
273
274 # output current progress indicator
275 my $last_p = 0;
276 sub progress {
277 #return if (! $opts{q}); # FIXME
278 my $current = shift;
279 my $total = shift || 1;
280 my $p = int($current * 100 / $total);
281 if ($p != $last_p) {
282 printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p );
283 $last_p = $p;
284 }
285 }
286
287 my $fake_dir = 1;
288 sub fakeprogress {
289 my $current = shift @_;
290
291 my @ind = ('-','\\','|','/','-','\\','|','/', '-');
292
293 $last_p += $fake_dir;
294 $fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0);
295 if ($last_p % 10 == 0) {
296 printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]);
297 }
298 }
299
300 # now read database
301 print STDERR "using: $type...\n";
302
303 if ($type_base eq "isis") {
304
305 my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!";
306
307 $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
308 my $db = OpenIsis::open( $isis_db );
309
310 my $max_rowid = OpenIsis::maxRowid( $db );
311
312 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
313
314 my $path = $database;
315
316 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
317 my $row = OpenIsis::read( $db, $row_id );
318 if ($row && $row->{mfn}) {
319
320 progress($row->{mfn}, $max_rowid);
321
322 my $swishpath = $path."#".int($row->{mfn});
323
324 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
325 $xml = $cp2utf->convert($xml);
326 use bytes; # as opposed to chars
327 print "Path-Name: $swishpath\n";
328 print "Content-Length: ".(length($xml)+1)."\n";
329 print "Document-Type: XML\n\n$xml\n";
330 }
331 }
332 }
333 print STDERR "\n";
334
335 } elsif ($type_base eq "excel") {
336 use Spreadsheet::ParseExcel;
337 use Spreadsheet::ParseExcel::Utility qw(int2col);
338
339 $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage);
340 my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!";
341
342 my $sheet = x($config->{sheet}) || die "no sheet in $type.xml";
343 my $start_row = x($config->{start_row}) - 1 || die "no start_row in $type.xml";
344
345 my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'";
346
347 my $sheet_nr = 0;
348 foreach my $oWks (@{$oBook->{Worksheet}}) {
349 #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n";
350 last if ($oWks->{Name} eq $sheet);
351 $sheet_nr++;
352 }
353
354 my $oWorksheet = $oBook->{Worksheet}[$sheet_nr];
355
356 print STDERR "using sheet: ",$oWorksheet->{Name},"\n";
357 defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file";
358 my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow};
359
360 for(my $iR = $start_row ; defined $end_row && $iR <= $end_row ; $iR++) {
361 my $row;
362 for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
363 my $cell = $oWorksheet->{Cells}[$iR][$iC];
364 if ($cell) {
365 $row->{int2col($iC)} = $cell->Value;
366 }
367 }
368
369 progress($iR, $end_row);
370
371 # print "row[$iR/$end_row] ";
372 # foreach (keys %{$row}) {
373 # print "$_: ",$row->{$_},"\t";
374 # }
375 # print "\n";
376
377 my $swishpath = $database."#".$iR;
378
379 next if (! $row);
380
381 if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
382 $xml = $cp2utf->convert($xml);
383 use bytes; # as opposed to chars
384 print "Path-Name: $swishpath\n";
385 print "Content-Length: ".(length($xml)+1)."\n";
386 print "Document-Type: XML\n\n$xml\n";
387 }
388 }
389 } elsif ($type_base eq "marc") {
390
391 use MARC;
392
393 $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
394 my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
395
396 # optional argument is format
397 my $format = x($config->{format}) || 'usmarc';
398
399 print STDERR "Reading MARC file '$marc_file'\n";
400
401 my $marc = new MARC;
402 my $nr = $marc->openmarc({
403 file=>$marc_file, format=>$format
404 }) || die "Can't open MARC file '$marc_file'";
405
406 my $i=0; # record nr.
407
408 my $rec;
409
410 while ($marc->nextmarc(1)) {
411
412 # XXX
413 fakeprogress($i++);
414
415 my $swishpath = $database."#".$i;
416
417 if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) {
418 $xml = $cp2utf->convert($xml);
419 use bytes; # as opposed to chars
420 print "Path-Name: $swishpath\n";
421 print "Content-Length: ".(length($xml)+1)."\n";
422 print "Document-Type: XML\n\n$xml\n";
423 }
424 }
425 } elsif ($type_base eq "feed") {
426
427 $import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);
428 my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!";
429
430 print STDERR "Reading feed from program '$prog'\n";
431
432 open(FEED,"feeds/$prog |") || die "can't start $prog: $!";
433
434 my $i=1; # record nr.
435
436 my $data;
437 my $line=1;
438
439 while (<FEED>) {
440 chomp;
441
442 if (/^$/) {
443 my $swishpath = $database."#".$i++;
444
445 if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) {
446 $xml = $cp2utf->convert($xml);
447 use bytes; # as opposed to chars
448 print "Path-Name: $swishpath\n";
449 print "Content-Length: ".(length($xml)+1)."\n";
450 print "Document-Type: XML\n\n$xml\n";
451 }
452 $line = 1;
453 $data = {};
454 next;
455 }
456
457 $line = $1 if (s/^(\d+):\s*//);
458 $data->{$line++} = $_;
459
460 fakeprogress($i);
461
462 }
463 }
464 }
465
466 # call this to commit index
467 $index->close;
468
469 1;
470 __END__
471 ##########################################################################
472
473 =head1 NAME
474
475 isis2xml.pl - read isis file and dump XML
476
477 =head1 DESCRIPTION
478
479 This command will read ISIS data file using OpenIsis perl module and
480 create XML file for usage with I<SWISH-E>
481 indexer. Dispite it's name, this script B<isn't general xml generator>
482 from isis files (isis allready has something like that). Output of this
483 script is tailor-made for SWISH-E.
484
485 =head1 AUTHOR
486
487 Dobrica Pavlinusic <dpavlin@rot13.org>
488
489 =head1 COPYRIGHT
490
491 GNU Public License (GPL) v2 or later
492
493 =head1 SEE ALSO
494
495 SWISH-E web site at http://www.swish-e.org
496
497 =cut

Properties

Name Value
cvs2svn:cvs-rev 1.28
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26